http://www.3dfx.ch/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=27132&g2_serialNumber=2

http://www.3dfx.ch/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=27136&g2_serialNumber=2

http://www.3dfx.ch/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=27139&g2_serialNumber=2

http://www.3dfx.ch/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=27142&g2_serialNumber=2

http://www.3dfx.ch/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=27145&g2_serialNumber=2

http://www.3dfx.ch/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=27148&g2_serialNumber=2

The board has a different bios then the "IBM 16MB with LCD" Instead this one is bios 1.06 and works perfect with the first motherboard I tried (unlike the IBM board which would not play nice with any BX board I tried.

Oh, and the IBM boards are clocked at 143 mhz, this one scoots along at 166 mhz =)

Yes, I'll dump the bios for those who want to try flashing their IBM board, I'm going to hit my 16MB board this weekend with this bios and see how much it improves things.

Wow, this has to be one of the last cards 3dfx themselves ever made (alongside your Rampage) since the memory-chips were made just two weeks before 3dfx went bust. They are one week later than the PCB of your Rampage so the card itself could have been completed even later than the Rampage which PCB-date is 0047...

The PCB date on this board is 4600, I suspect it was a later card than the IBM cards based on the bios revision, clock speed and robust operation even though the PCB is the same.

And of course the 4800 ram chip dates.

I dumped the bios for anyone who wants it.

Tried flashing my two 16MB boards this morning using my regular BX testbed, lol.

The machine will not boot up at all with the 16MB boards installed, works perfect with the 32mb board.

Looks like I have to go up in the attic at dust off the old Pentium I setup with TX chipset, it was the only thing I got to work with the 16mb boards and flash them with that.

Both boards flashed ok, still no function on the BX chipset though.

Something is different on a hardware level than the 32 mb board.

Oh, the "production" A1 16mb board flashed without a problem but the earlier A0 displayed a warning message stating hardware wasn't on file, still flash?

Curious.

okay

my 16MB daytona also dont work on a BX mobo (asus p2b-ds)

the only other mobo i tried is a Asus a7n8x-vm400 (nForce2 IPG) and works perfect
the picture errors came from the card itself
http://www.tdfx.de/pic/voodoo4_4200pci_16mbsg_52.JPG
one lane is broken near R504

where is the bios download?

wrote on 24.11.07 at 16:17:04:

where is the bios download?

In your "inbox"

=)

I doubt the board will work on a BX like mine didn't, but let us know please if it does.

thx for the BIOS file

my card dont work an a BX mobo
and works on a A7N8X-VM400
other mobos i dont test

Digging around I have found some interesting things.

1) The 16MB boards sold on Ebay were indeed from IBM UNIX servers, as strictly 2D display devices.

2) Poking around the A0 bios, it is clearly not for a regular PC and even has some markings related to IBM and Unix systems.

3) Downclocking the 1.06 bios to 100 mhz has my A0 board up and functioning on the Pentium TX platform with only minor graphics glitches. Looks like it has bad ram but the reduced clock speed helps as the display has gone from almost impossible to almost normal.

4) I suspect that the "IBM" boards have some changes to them related to running on UNIX hardware, perhaps a board level look of both cards will provide a clue.

I suspect I have the only Daytona (for the moment) with this PCB configuration that runs without issue on most motherboards.

Quote:

one lane is broken near R504

@exxe - can you highlight it on the picture...I can see R504, but not the broken trace very clearly!

If anyone wants me to test anything let me know, I still have the TX (200mmx cpu & Win98) platform setup.

http://www.abload.de/thumb/pict0289shs.jpg

WOW! Someone really scraped that!

The trace is cut in 3 places.

But it's an easy fix...

1) Scrape the green coating off the VIA at the end of the trace.
2) Solder a wire to the gold VIA at the start of the trace.
3) Solder the other end of the wire to the VIA you scraped.

FIXED!

Be sure the wire is as short as possible and that its good quality wire.

i dont solder a wire on it

maybe some "silberleitlack" Electrical conductivity Lacquer dont know the right english word
and do it like the old Athlon XP bridgetrick




Quote:

Be sure the wire is as short as possible...

hmm the trace itself isnt as short as possible

Very odd- Did some testing with the 16mb vs 32mb board and I present some very odd results.

Running both glquake and quake II I get the exact same framerate.

46 fps in glquake
20 fps in quake II

Picture quality is perfect in quake II but somewhat garbled with the 16mb board under glquake. I suspect the issue is the 16mb board reporting it has 32mb but at a loss to explain the same fps in both games.

The cpu should not be a huge factor in glquake (I had AA on and triple buffering to really stress the video board) with the res set at 1024 x 768 in both games.

Thoughts?

System is a 200mmx with 256 mb of ram and Win98se.

Running 3dmark2000(v11) the 32 mb board will generate a score of 475, however the 16 mb card doesn't finish it chokes on one of the final tests (runs out of ram I bet)

Anything other tests before I put the system back away?

Is the chip id different from other Daytonas you have/have had?

I mean the '355-0028-400' and '0046 2000' numbers.

BTW, does it have FSAA capabilities? If so, a comparison between a V4 and a Daytona would be nice. Maybe 3dfx improved it in the Daytonas :)

wrote on 26.11.07 at 07:48:42:

Is the chip id different from other Daytonas you have/have had? I mean the '355-0028-400' and '0046 2000' numbers. BTW, does it have FSAA capabilities? If so, a comparison between a V4 and a Daytona would be nice. Maybe 3dfx improved it in the Daytonas :)

1) I don't think so but will double check.

2) yes, aa works perfect.

3) I did toss in a V4-4500 PCI last night and ran Mark2000 again, the V4 scored higher than the Daytona typically 490 vs the Daytona's 475.

I suspect the new ram controller or the ram settings.

Quote:

3) I did toss in a V4-4500 PCI last night and ran Mark2000 again, the V4 scored higher than the Daytona typically 490 vs the Daytona's 475.

I meant, any difference in FSAA? :)

Gary,

can u please test the card in a modern PC ?
Please also post a screenshot from Everest Tool, to see the GPU Properties ;)

Br.
Radu.

wrote on 26.11.07 at 17:56:58:

Gary, can u please test the card in a modern PC ? Please also post a screenshot from Everest Tool, to see the GPU Properties ;) Br. Radu.

I can't, the only platform the 16mb board will work here is the old TX system (aside from the 32mb card which works in everything)

The Daytona is just a VSA-100 with a smaller die and DDR controller.

The VSA-100 runs 128 bit SDRAM, the Daytona 64 bit DDR ram.

Everest just states it is a "VSA-200" and gives nothing for GPU properties.

Benches with a 143 mhz Daytona are here=

http://www.thedodgegarage.com/3dfx/bench.htm

Hmm... maybe it is time to do them again...

wrote on 26.11.07 at 17:56:58:

Gary, can u please test the card in a modern PC ? Please also post a screenshot from Everest Tool, to see the GPU Properties ;) Br. Radu.

+1 on the modern machine test 8)